Wavevector multiplexed atomic quantum memory via spatially-resolved single-photon detection

Abstract

Parallelized quantum information processing requires tailored quantum memories to simultaneously handle multiple photons. The spatial degree of freedom is a promising candidate to facilitate such photonic multiplexing. Using a single-photon resolving camera, we demonstrate a wavevector multiplexed quantum memory based on a cold atomic ensemble. Observation of nonclassical correlations between Raman scattered photons is confirmed by an average value of the second-order correlation function [Formula: see text] in 665 separated modes simultaneously. The proposed protocol utilizing the multimode memory along with the camera will facilitate generation of multi-photon states, which are a necessity in quantum-enhanced sensing technologies and as an input to photonic quantum circuits.

Fig. 1

Proposal for single-photon spatial routing and multiplexing for multi-photon state generation. A single-photon resolving camera registers consecutive frames during the quantum memory write-in process. Each detection heralds creation of a spin-wave excitation in the atomic ensemble quantum memory with a wavevector determined by the position (k _xi, k _yi) at which i-th photon was registered. Acquisition and write-in continue until the desired number of excitations has been created. At this time a photonic switch is reconfigured to channel photons from conjugate directions stored in a classical memory and subsequently the readout pulse is applied to convert stored spin-wave excitations to the requested number of photons, which will be used later, e.g., in the quantum circuit

Fig. 2

Experimental realization of the wavevector multiplexed quantum memory. a Schematic of the main part of the experimental setup. The atomic ensemble released from the magneto-optical trap (MOT) is illuminated with orthogonal circularly polarized write and read laser beams. Angles at which Stokes (S) and anti-Stokes (AS) photons (produced through Raman scattering) are emitted from the atomic ensemble are imaged on the single-photon resolving I-sCMOS (intensified scientific complementary metal-oxide semiconductor) sensor, composed of a sCMOS camera and an image intensifier. Optically pumped atomic cells (S and AS filters) filter out the residual laser light and stray fluorescence. b Example subsequent frames in the Stokes (bottom) and anti-Stokes (top) regions demonstrating correlated photon pairs in each camera frame. Note that while most frames will contain no photon or a photon only in a single region, almost all (>90%) frames with a coincidence event will contain a correlated photon pair for the detection probability of S photon p _S = 1.2 × 10⁻². c Pulse sequence used in the experiment (see Methods for details) consists of trapping magnetic field switching, laser cooling, and optical pumping (with depletion) preparation stages, as well as short write-in and readout laser pulses (quantum memory stage) producing Stokes and anti-Stokes photons. d Atomic level configuration (colors correspond to the pulse sequence in c). In the process of write-in and readout, the spin wave (green arrow) is created and annihilated, respectively. The wavy arrows correspond to S (blue) and AS (purple) photons

Fig. 3

Spatial properties of the generated biphoton state. a All Stokes–anti-Stokes coincidences obtained from 10⁷ frames marked with their k _y wavevector components (k _x for the inset) for zero memory storage time, demonstrating high degree of momenta anti-correlation. Each plot disregards the perpendicular component. b Same coincidences counted in the center-of-mass variables (k _x,S + k _x,AS) and (k _y,S + k _y,AS). The central peak is fitted with a two-dimensional Gaussian to obtain its center and width. One-dimensional distributions correspond to cross-section counts selected for central pixels. Both in a, b, neither accidental nor noise background subtraction is performed

Fig. 4

Nonclassical correlations of photons emitted from the quantum memory. a, b Second-order cross-correlation function $g_{\mathrm{S,AS}}^{\left(2\right)}$ measured for different positions of ROI in S and AS arms, for zero memory storage time. Nonclassical correlations are observed only between conjugate modes, confirming the highly-multimode character of our quantum memory. Data corresponds to S photon probability p _S = 2 × 10⁻⁴ per ROI. Standard deviation error maps are included as Supplementary Fig. 1. c $g_{\mathrm{S,AS}}^{\left(2\right)}$ for Stokes and anti-Stokes photons measured at t = 0 storage time using different sizes of ROI in the analysis. Smaller ROIs correspond to lower p _S and consequently give higher values of $g_{\mathrm{S,AS}}^{\left(2\right)}$. Our theoretical prediction for $g_{\mathrm{S,AS}}^{\left(2\right)}$ calculated for the measured mode size closely adheres to experimental results (see Methods for details). Other curves correspond to the maximum value of $g_{\mathrm{S,AS}}^{\left(2\right)}$ without noise in the AS arm and the maximum theoretical result for two-mode squeezed vacuum state (TMSV) with given probability p _S. Gray dashed lines mark the regime of operation used in the measurement shown in a, b. d Second-order correlation as a function of storage time, measured for two different angles of scattering corresponding to stored spin waves with different K _x. Data was taken with a higher than in (a–c) S photon detection probability of p _S = 1.9 × 10⁻³ and thus the value of the correlation function is smaller. Nonclassical correlations for spin waves with smaller wavenumber are confirmed for the storage time t up to 50 μs. Theoretical model of the time evolution of $g_{\mathrm{S,AS}}^{\left(2\right)}$ (solid lines, see Methods for derivation) exhibits good agreement with experimental data, except for the initial drop that we attribute to an increase of noise fluorescence of thermal atoms. Errobars in c, d correspond to one standard deviation drawn from an ensemble of multiple conjugate region pairs

Fig. 5

Characterization of the atomic filtering system for Stokes photons. Data points correspond to a measurement of absorption of the weak probe beam while fit is the theoretical prediction of OD based on a Voigt-profile absorption model including optical pumping. Inset shows the OD in a larger scale (only the fitted function), demonstrating very high attenuation of write laser light (ν − ν ₀ = 4.3 GHz detuning). Simultaneously, high transmission of S photons (ν − ν ₀ = −2.5 GHz detuning) is achieved. Similar characteristics were also obtained for the AS filter. Detuning is given with respect to the line centroid at ν ₀. Two more central absorption peaks corresponds to a residual amount of ⁸⁵Rb in the filtering glass cell. Experimental errorbars correspond to one standard mean error derived from many collected spectra

Fig. 6

Result of the eigenmode decomposition for the number of modes. Dots represent the results from numerical decomposition, whereas the solid line is the simplified prediction of 0.565κ/2σ. Dotted gray lines correspond to values of σ _x,y we obtain in our experimental setup and corresponding numbers of modes M _x and M _y